Modeling Demand Module 7 Conceptual Structure of SIMQ Market Model Firm Demand = Total Industry Demand * Share of Market Firm Demand = Average Firm Demand * n * Share of Market Firm Demand = Average Firm Demand * Normalized Share of Market Macro-economic Influences Seasonal Patterns Exogenous Demand Stage of Life Cycle Industry Activity Pricing, Promotion, Endogenous AFD Demand FD Quality Competitive Profile Relative Pricing, Promotion, Quality, and Loyalty Relative Demand NSOM Normalised Share of Market NSOM = Firm Demand / Avg Firm Demand Relative Price (current) Relative Advertising (current, t-1, t-2 ) Pricing NSOM Promotion Relative R&D (t-1, t-2) Loyalty NSOM (t-1) Quality NSOM is firm specific and a measure of relative demand, the predictor variables should also be relative to industry averages. For example, relative price of the firm is PREL = Firm’s Price / Industry Avg. Price Calculating NSOM 11) Use a multiple regression to estimate NSOM. Average Firm Demand How many units will any firm sell on the average. D n Df f 1 Error n Endog Exog AFD = Exogenous demand + Endogenous demand Exogenous demand = “Base demand” X Seasonal effects Endogenous demand = Influence of aggregate industry behavior Average Firm Demand AFD Exogenous Demand Macro-Economic Influences Endogenous Demand Industry Behavior - Pricing (Avg Price) - Seasonality - Promotion (Avg. Advertising) - Stage of Life Cycle - Product Quality (Avg. R&D) Estimate Trend and Seasonality using Time Series Analysis Estimate weights of these factors using Regression Analysis AFD = {(T*S) + (B0+B1*Avg P+..)} AFD: Exogenous Demand Base demand Population, Income, Tastes, Product life cycle, Substitutes and complements (Macro-economic influences) Seasonal demand Weather, Customs, Holidays Not all products are affected Estimation is done using: Time Series Decomposition Calculating Exogenous AFD 1) Use Statpro to generate Seasonal Indices Calculating Exogenous AFD 2) Use Seasonal Indices to De-seasonalize observations. 1438 / .895 = 1607 Calculating AFD 3) Fit a simple regression line to the de-seasonalized observations. Calculating AFD 4) Use the regression line to create a ‘de-seasonalized’ forecast. 11 * 65.213 + 1237.9 = 1955.2 Calculating AFD 5) Re-seasonalize the predicted forecast. This is the exogenous portion of demand. 1.093 * 2150.9 = 2350.3 Calculating AFD 6) Calculate the residual error from the forecast. This is the endogenous portion of demand. 2839 – 2350.3 = 488.7 Calculating AFD 7) Fit a multiple regression with the residuals as the dependent variable and the firm data as the independent variables. Calculating AFD 8) Use the new regression equation to forecast residuals. These are estimated of endogenous demand. 15688.25 + (-43.885) * 371.9 + (.0032) * 97960 + (.0127) * 25610 = 4.2 Calculating AFD 9) Add the exogenous forecast to the endogenous forecast to create a composite forecast. 2350.3 + 204.2 = 2554.6 Calculating AFD 10) Subtract the composite forecast from the observations to find a total error. 2839 – 2554.6 = 284.4 Calculating AFD 11) Calculate the standard deviation of the residuals. Because there may be more explanation that can be squeezed out of the trend and attributed to the dependent variables, repeat the process. For subsequent iterations, Raw AFD – Estimated Endogenous as starting data for time series. Continue till error stops decreasing. Conceptual Structure of SIMQ Market Model With Average Firm Demand and Normalized Share of Market modeled, we can now create a decision support system for any individual firm. Exogenous Demand Endogenous AFD Demand Relative Demand FD NSOM Example DSS